INTELLIGENCE/OTTRED HOT TAKES
OTTRED HOT TAKESPOLL + ANALYSIS8 MIN READ

THREE POLLS, 65 RESPONSES, AND NOT ONE VOTE FOR OVER-HYPED

THE AD IS
FREE NOW.
THE AUDIENCE
ISN'T.

A convincing thirty second ad used to take six weeks, a crew, a colourist, a sound house and a clearance lawyer. This morning it takes minutes and a licence. We put the same question to LinkedIn and to the OTTRED community. The market backed new formats at 39 per cent. Forced to one answer, the operators said the market gets flooded with noise, 66 per cent. And nobody, in any poll, anywhere, said the whole thing was over-hyped.

FILED BY

OTTRED Intelligence

PUBLISHED

NOVEMBER 2026

FEATURING

LINKEDIN + COMMUNITY POLLS

A creative director alone at 2am facing a wall of screens running different versions of the same commercial

THE EDIT SUITE, 2AM. TWELVE COMPETENT VERSIONS AND NOBODY LEFT TO ARGUE WITH.

66%

of the OTTRED community said it floods the market with noise, forced to one answer

0

votes for less than the hype, when the community were given free choice

39%

on LinkedIn still backed it to unlock formats we can't imagine yet

65

responses across three polls, run in the same week in November 2026

01 — The Question

AI can now produce a convincing 30 second ad in minutes. Within five years, what does it do to our industry?

For most of the last fifty years the thirty second ad was a small film. It had a director, a casting session, a location, a wardrobe rail, a food stylist and a lawyer who read the script twice. It took six weeks and it was, in the way that mattered to everybody involved, expensive. Expense was not a byproduct of the process. It was the process. It was the reason the idea had to be argued for before anyone was allowed to shoot it.

That is now optional. Through 2024 and 2025 the tools moved out of the demo reel and into the places where media actually gets bought. Veo 3 arrived inside Google Ads with audio and eight second clips. Meta put generative creative in the same window where its advertisers already build campaigns. Sora 2 made photoreal ten second video a consumer product and then had to invent rights rules in public, because the first thing everyone did was point it at somebody else's work.

The numbers underneath went boring, which is how you know a technology has won. Meta has said more than a million advertisers have used its generative creative tools, with tens of millions of ads produced, and its automated campaign layer is running at a scale that would place it among the largest media owners on earth on its own. Nobody ran a campaign about that. It simply became the default, quietly, for the advertisers least likely to be invited onto a panel about it.

Which brings us to the actual scarcity. The world has never lacked advertising minutes, and it does not lack them now. It lacks attention, and attention is the one input that did not get a model upgrade this year. If the cost of a competent frame falls towards zero while the supply of attention stays fixed, the industry's problem stops being production and becomes entirely a question of who is worth watching. That is not a technology question. It is a strategy question, and it is being answered badly in most boardrooms.

So we asked the only question with an interesting answer left in it. Not what can it do. What does it do to us. Same wording to LinkedIn and to the OTTRED community, one vote each on LinkedIn, five options including Other in the community, followed by a second community poll with multiple selections allowed, because forcing a single answer on a subject like this hides the thing we most wanted to see.

A note on method. Eighteen votes on LinkedIn, thirty two single select votes and fifteen selections in the OTTRED WhatsApp community. This is not a representative sample of the global advertising industry. It is a room of people whose job this is, and a room of people watching it from outside, and the distance between the two rooms turns out to be the entire story.

Competence is about to become the cheapest thing in the building. The only question worth asking after that is what you are selling once it is.

02 — The Results

The same question, three ways

LinkedIn got four options and one vote each. The community got five, including Other. Then we ran it again as select all that apply, because a room that disagrees with itself is more useful than one that doesn't.

LinkedIn, single choice, 18 votes

4 options, no Other

Unlocks formats unimaginable
39%
Makes great content cheaper
28%
Floods the market with noise
28%
Less than the hype suggests
6%

The market split three ways and picked the optimistic answer by a nose. 39, 28, 28 is not a verdict. It's three different jobs answering one question. The creative director sees possibility, the media buyer sees cost, the performance marketer sees inventory. On LinkedIn everybody is right and nobody is certain.

The Gap

Where the market and the room disagree

The percentages matter less than the distance between them. On noise the two audiences are 38 points apart. On new formats they are 17 points apart in the opposite direction, and on over-hype they are three points apart at almost nothing.

LINKEDIN

OTTRED COMMUNITY

28%
Floods the market with noise
66%

38 points apart

39%
Unlocks formats unimaginable
22%

17 points apart, the other way

28%
Makes great content cheaper
6%

22 points apart

6%
Less than the hype suggests
3%

the only answer nobody backed

03 — The Finding

The only answer nobody gave

Less than the hype suggests. Six per cent on LinkedIn. Three per cent in the community. Zero, absolute zero, among the people who were allowed to pick more than one thing. Across sixty five responses, roughly one person in any given room believes AI ad production is over-hyped.

That single number reframes every other number in this piece. The industry is not having a debate about whether this is real, and anyone still running that argument is running it a year behind the market. The capability has been conceded. What is being fought over now is consequence: whether a tool that makes competence free makes the work better or simply louder, and who gets paid when the answer turns out to be neither.

It also explains the apparent contradiction in our own results. The same room that said noise by two thirds, given the chance to hold two ideas at once, went straight back to new formats. That is not confusion. It is the most accurate answer in the dataset, and the only one worth building a plan on.

6%

LinkedIn said less than the hype

3%

Community, forced to one answer

0%

Community, free to pick anything

An archive wall of thousands of near identical generated ad thumbnails, one screen lit bright
Nobody voted against the technology. They voted against what a market looks like when every competent frame is free to make.

04 — How We Got Here

Four years, in four moments

2022 — 2023

The models arrive, and Firefly makes it a procurement question

DALL·E 2, Midjourney and Stable Diffusion turn comps into a same-day exercise. Then Adobe ships Firefly trained on licensed content and sells it as commercially safe, which is the moment legal rather than creative starts paying attention.

FEB 2024

Sora is unveiled, and production companies start building AI units

Text to video stops being a novelty demo. The phrase craft quietly becomes a negotiating position rather than a compliment.

NOV 2024

Coca-Cola remakes its own Christmas ad, and the reaction is ferocious

Technically seamless, publicly rejected. The industry learns that audiences judge craft and credit long before they judge technology.

2025 — NOW

The toolchain moves into the buying interface, and rights become the argument

Veo 3 lands inside Google Ads with audio. Meta scales generative creative past a million advertisers. Sora 2 goes consumer, and the fight turns to likeness, licensing and every archive deal signed before 2024.

05 — What The Market Is Seeing

Six things that actually happened, and what each one told us

This argument is not theoretical. It has been running in public for two years, and every stage of it left a mark you can point at.

NOV 2024

Coca-Cola remakes its own Christmas ad with AI

Coca-Cola rebuilt its 1995 Holidays Are Coming truck commercial, one of the most recognisable ads ever made, with AI production partners, Secret Level widely named among them. The ad ran. It also detonated. Animators, VFX artists and ad directors objected in public, and Coca-Cola's answer was measured: it will keep using AI where it works.

The capability was never in question. The argument was craft, credit and who gets paid, and it started the moment the audience could sense the difference without being able to name it.

2023 TO 2025

Guess puts a synthetic model in Vogue France

Guess ran a campaign built around an AI-generated model in Vogue France. Levi's had tried AI-generated models two years earlier and met a different storm. Both were technically seamless. Neither argument was about image quality.

When the synthetic thing is a person, the public conversation stops being about the technology and starts being about representation, work and who is allowed to be in the picture.

2025

Meta quietly industrialises the whole thing

While the industry argued, the plumbing got installed. Generative creative sits inside Ads Manager. Meta has said more than a million advertisers have used its generative tools, with tens of millions of ads produced, and its automated campaign layer Advantage+ is running at roughly a 60 billion dollar annual rate.

Nobody voted on this. It is simply where a very large share of the world's ad creative now gets made, by the advertisers least likely to be invited onto a panel about it.

2025

Google puts the camera inside the dashboard

Veo 3 arrived in Google Ads with audio, eight second clips and variants on tap, positioned almost casually around the ad you couldn't afford to shoot. AI Max in Search did the same to targeting and copy. The ad stopped being a thing you make and became a setting you turn on.

The toolchain is now inside the buying interface. That matters more than any single model release, because it means production decisions get made by whoever is sitting in the campaign build.

AUTUMN 2025

Sora 2 moves the argument to ownership

OpenAI's Sora 2 put photoreal ten second video into everyone's hands, and within days the fight had moved from what the model can do to who owns what it makes. Rights holders were given control over likeness, and studios and brands started asking what happens when their presenter, their logo and their house look are all just prompt text.

If you can generate the ad, you can generate somebody else's ad. Every talent and archive contract signed before 2024 is now a gap with a delivery date.

2025 TO 2026

The agency layer consolidates around it

Omnicom closed its purchase of IPG in November 2025 to build the largest agency group in the business. WPP cut guidance more than once, with AI driven shifts in client spending and restructuring in the mix. Production companies and VFX houses felt it first: the mid market shoot, the four week edit, the day rate for a tabletop sequence.

Consolidation at the top and automation in the middle is the same story told from two ends. The squeeze lands on the layer that used to convert briefs into craft.

06 — The Arithmetic

What a thirty second ad actually was

The industry talks about this as a cost story. It isn't. It's a headcount story. Six stages collapsed into three, and the two that vanished were the two that employed the most people.

THE OLD CHAIN

Six weeks, and a queue of people who all get paid

  1. 01Brief, then a round of concepting
  2. 02Pre-production: director, casting, locations, wardrobe
  3. 03Shoot days, crew, kit, insurance
  4. 04Edit, colour, sound, music clearance
  5. 05Legal, compliance and clearance sign-off
  6. 06Versioning for every market, format and ratio

Collapsed

THE NEW CHAIN

An afternoon, and one decision that actually matters

  1. 01Prompt, reference frames, a licence
  2. 02Twelve generated versions before lunch
  3. 03A human chooses, corrects and signs off
An empty film soundstage at night with the lights still warm and a laptop glowing on an apple box
The room is still there. The queue of people who used to fill it is the part that has gone.

The old chain had a quality control built into it, and it wasn't the client. It was the number of people who had to agree that the idea was worth spending six weeks on.

07 — Who Wins, Who Pays

Six parties, and nobody leaves with everything

The polls asked what happens to the industry. This is that question split into the groups who will actually feel it.

Brands

GAINSSpeed, and the ability to test ideas that were never affordable to shoot. Ten variants where one used to exist, in every market, in every dialect.

GIVES UPThe scarcity that made an ad feel expensive. When everyone can make a competent one, being noticed is the only differentiator left.

Agencies

GAINSIdeas can be shown rather than described. The pitch moves, which is worth real money in a room where the client has seen thirty decks this quarter.

GIVES UPThe production margin and the retainer hours that paid for it. The consolidation at the top of the holding groups is the sound of that adjustment.

Production and VFX

GAINSGenuinely new work: clean-up, set extension, localisation, de-ageing, versioning. Invisible, technical, and in more demand than ever.

GIVES UPThe mid-market shoot and the four week edit. The reliable middle of the business is always what disappears first.

Platforms

GAINSBoth sides of the trade at once, the tool that makes the creative and the inventory it fills. More supply means more auctions, and volume is the business model.

GIVES UPNothing structural. Distinctiveness is the advertiser's problem, and the auction clears whatever the market brings.

Talent and crews

GAINSNew categories of work, and the option to license a likeness rather than lose a day rate entirely.

GIVES UPLeverage, in a negotiation where the alternative is a generator. Every contract that predates 2024 needs revisiting, and most have not been.

Audiences

GAINSMore relevant, better localised, and occasionally more inventive advertising than the same six ideas shot again in a different kitchen.

GIVES UPThe signal. When everything is competent, attention gets more expensive for everyone, including the people doing the watching.

Read down the middle column and the pattern is uncomfortable but clear: the parties capturing most of the released value own either the toolchain or the inventory. Everyone else is negotiating over the remainder.

08 — What It Means

Five things to plan around, not five things to feel

If everybody already accepts the capability, the only useful position is a specific one. These are the five consequences we would be acting on this quarter, whichever side of the poll you sat on.

01

The scarce skill moves from making to judging

When anyone can produce twelve versions by lunchtime, the value sits in knowing which one is right and being able to say why in a room full of people who all have twelve of their own. Taste, defended with evidence, becomes the job.

02

Distinctiveness stops being a nicety and becomes the strategy

Competent is now the floor, and floors don't get remembered. The brands that pull away will be the ones whose assets are recognisable with the sound off, at three seconds, in a feed full of things that look like everything else.

03

Likeness and rights become board level

If a model can produce your presenter, your deal needs an AI clause, and so does your music, your archive and your talent. The brands caught out will be the ones who treated this as a legal tidy up rather than a commercial term.

04

The buying side consolidates while the making side fragments

Two enormous groups at the top, millions of small makers at the bottom, and a thinner middle than the industry has ever carried. Career planning should assume that shape rather than the old pyramid.

05

The regional gap widens before it closes

The toolchain is priced and distributed in dollars, and in markets where senior seats are already thin, the temptation is to automate the junior work that was the training ground. That is the risk nobody is polling on, and it will show up in five years as a missing generation.

09 — The Arguments Still Being Made

Three claims worth retiring

Every one of these is said in good faith, usually by someone brilliant. None of them survive contact with what has actually happened since 2024.

It's cheaper, so it's worse.

The evidence is narrower than either side claims. Audiences punish visible AI in front of the camera, a face or a body, and barely register it behind the camera: clean-up, extension, localisation, versioning. Almost all of the AI in the ads you saw this month was invisible, and that is exactly where the money moved first.

Nobody can tell the difference.

They can, and they say so, loudly. Coca-Cola's remake was technically seamless and still produced a public argument about craft and credit. The tell is rarely the pixels. It is the absence of a decision, and people feel that long before they can name it.

The hype will pass.

Sixty five responses across three audiences, and the option nobody chose was less than the hype. The people with budgets to lose have already stopped calling it hype, which is the most reliable signal this industry ever gives.

10 — The Regional Read

The Gulf read, because we get asked most

Where the cost of production collapses unevenly, the commercial advantage moves to whoever already owns the thing nobody can generate.

A Gulf city skyline at dusk with a huge vertical LED screen running a generated luxury ad
SAME MODEL, FIVE DIALECTS, ONE AFTERNOON. THE BARRIER THAT PROTECTED REGIONAL PRODUCTION IS NOW THE THING THAT EXPOSES IT.

Arabic is not one market. Dialect, humour and idiom shift within a few hundred miles, and localisation was the reason a campaign never travelled further than the city it was shot in. When a model can re-voice and re-cut a spot in five dialects before the client calls back, the barrier that protected regional production houses becomes the thing that exposes them.

The advantage that survives is the one that was never for sale. Knowing which version lands in Jeddah and which gets laughed at in Beirut is still the product. Production capacity was only ever the packaging around it, and packaging just got free.

The risk is structural and identical to the one we flagged in Monday's Role Finder. In a region where senior seats are already thin, the temptation is to automate the junior work that was the training ground for everyone currently in charge. Cheaper output this year, and a missing generation in five.

The opportunity is just as real. Brands in this market have never been able to afford the volume of creative that the US and Europe treat as routine, and that gap is now a budget conversation rather than a capability one. The people who will win it are the ones who can brief precisely, in a room full of generated options, in a language they actually speak.

11 — What To Watch

Seven signals that will settle this argument

None of them are about model quality. All of them are about money, ownership and attention, which is where this was always going to be decided.

  1. 01

    Whether platforms start labelling or grading AI creative, and what that does to the price of a generated impression.

  2. 02

    Whether brand lift and attention data starts splitting by production method, because that is the moment the argument becomes commercial.

  3. 03

    Whether the mid market production company survives as a business, or becomes a freelance network attached to a platform.

  4. 04

    Whether the Gulf and Asia Pacific build their own AI content capability or import it, and who ends up owning the models trained on local work.

  5. 05

    What the first genuinely new ad format of this era actually looks like, and whether anyone can name the person who made it.

  6. 06

    Whether procurement starts asking agencies to price the toolchain into the fee, because that is when the production margin finally moves.

  7. 07

    Whether any brand is brave enough to run generated creative at the top of the funnel and prove it in a brand lift study, rather than burying it in performance.

12 — The Takeaway

Nobody thinks the tool is over-hyped. That is the whole story.

Sixty five responses, three polls, and not one audience willing to say this is more noise than substance. The market believes the capability. The room believes the consequence. The only thing they agree on is that the argument about whether it works is over, and it ended before most of the industry noticed it had begun.

What replaces it is less comfortable, because it cannot be settled by buying a licence or hiring a prompt engineer. If competence is free, then competence is worthless as a differentiator, and every brand, agency and freelancer in this business is about to be measured on the thing that was always hardest to fake: knowing what the work is for.

Which leaves one useful question for anyone whose job touches advertising this year. Not whether you use it. What is the ad for, and what should the machine be allowed to decide on its own. Get specific and the technology becomes an advantage. Stay vague and you join the eighty per cent of new inventory that nobody remembers seeing.

The winners in five years will not be the ones who adopted fastest. They will be the ones who were clearest about what they were making, and who could still explain why to a room that had twelve other options on the table.

OTTRED INTELLIGENCE

The room that answered this poll is the room you want to be in

OTTRED is the intelligence hub built for the global streaming, OTT and media technology industry. Join over 2,200 professionals across 91 countries and vote on the next one.

DOWNLOAD THE OTTRED APPMORE INTELLIGENCE
WWW.OTTRED.COM
Get in touch with our team